# Dynamic Parametric Evaluation Suite for LLM Reasoning Benchmarks

**URL:** <https://discuss.huggingface.co/t/dynamic-parametric-evaluation-suite-for-llm-reasoning-benchmarks/190982>\
**Category:** Research\
**Created:** [October 25, 2026, 2:27am UTC](https://discuss.huggingface.co/t/dynamic-parametric-evaluation-suite-for-llm-reasoning-benchmarks/190982 "2026-10-25T02:27:37Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![qurbi](https://avatars.discourse-cdn.com/v4/letter/q/f0a364/32.png) [@qurbi](https://discuss.huggingface.co/u/qurbi)\
**Post date:** [October 25, 2026, 2:27am UTC](https://discuss.huggingface.co/t/dynamic-parametric-evaluation-suite-for-llm-reasoning-benchmarks/190982/1 "2026-10-25T02:27:37Z")

</div>

Hi everyone,

I’ve been tracking the recent engineering developments around reasoning capabilities and symbolic logic. I wanted to share a benchmarking suite I recently built natively on the Hub to explicitly stress-test frontier models on multi-step analytical math and physics execution.

It programmatically targets structural failures where model gradients collapse or spatial constraints fail:

- Interactive Mechanics Playground: [Mechanics Parametric Evaluation Suite - a Hugging Face Space by qurbi](https://huggingface.co/spaces/qurbi/mechanics-parametric-evaluation-suite)
- Interactive Calculus Playground: [Calculus Parametric Evaluation Suite - a Hugging Face Space by qurbi](https://huggingface.co/spaces/qurbi/calculus-parametric-evaluation-suite)

I’d love to know if anyone in the research community is interested in using these parametric JSON suites to stress-test your upcoming model iterations or logic pipelines!